A column generation approach for power-aware optimization of virtualized heterogeneous server clusters

Increasingly, clusters of servers have been deployed in large data centersto support the development and implementation of many kinds of services,having distinct workload demands that vary over time, in a scalableand ecient computing environment. Emerging trends are utility/cloudcomputing platforms, where many network services, implemented andsupported using server virtualization techniques, are hosted on a sharedcluster infrastructure of physical servers. The energy consumed to maintainthese large server clusters became a very important concern, which inturn, requires major investigation of optimization techniques to improvethe energy eciency of their computing infrastructure.In this work, we present a novel optimization approach that simultaneouslydeals with (1) CPU power-saving techniques combined with serverswitching on/o mechanisms, (2) the case of server heterogeneity, (3) virtualizedserver environments, (4) an ecient optimization method basedon column generation techniques. The key aspects of our approach arethe basis on rigorous and robust optimization techniques, given by highquality solutions in short amount of processing time, and experimentalresults on the cluster conguration problem for large-scale heterogeneous server clusters that can make use of virtualization techniques.